Terror From Within: The Political Determinants of Domestic Terrorism 1

Transcription

1 Terror From Within: The Political Determinants of Domestic Terrorism 1 Sara Polo 2 Department of Government University of Essex March 14, 2012 (Draft, please do not circulate) Abstract How do domestic political institutions affect the likelihood of terrorism within a country? Existing studies mainly focus on transnational terrorism and rely on regime type as an aggregate concept. These studies contend that democracies are more prone to terrorism; yet, focusing on the opportunities provided by democratic regimes vis-à-vis non democratic ones has largely disregarded how certain domestic political institutions can generate grievances among sub-national groups who may then turn to terrorist violence. Given such an inconclusive picture, this study investigates the largely unexplored phenomenon of domestic terrorism and disaggregates regime types into specific institutions based on their effect on the openness and inclusiveness of the political system. The empirical analysis confirms the hypothesis that political discrimination of minority groups and ethnic exclusion increase the risk of domestic terrorism regardless of regime type. Moreover, countries characterised by intermediate levels of political openness are more likely to experience domestic terrorism. 1 Paper to be presented at the workshop on Advancing the Scientific Understanding of Conflict and Cooperation: Alternative Perspectives from the UK and Japan, Colchester, UK, March This paper is part of a PhD project on determinants and spatial patterns of domestic and transnational terrorism. Please note that this is work in progress, and therefore preliminary and incomplete. I acknowledge the financial support of the ESRC. 2 I thank Todd Sandler for kindly sharing his data. smtpol@essex.ac.uk. 1

2 Introduction (preliminary) This study investigates the causes of domestic terrorism and how these may differ from the causes of transnational terrorism. I will devise a systematic model of domestic terrorism at the country-level, which differentiates between direct causes of terrorism (i.e. factors that generate grievances) and permissive conditions, to understand how the characteristics and quality of a country s political environment influence the risk of domestic terrorism. This will be one of the first attempts to systematically analyse the political determinants of domestic terrorism and compare them with determinants of transnational terrorism. In fact existing quantitative research has almost exclusively focused on transnational terrorism and relied on regime type as an aggregate concept. It has found that democracies are more prone to transnational terrorism. However, looking at the organizational opportunities and strategic incentives provided by democratic regimes vis-à-vis non democratic ones has largely disregarded the fact that domestic political institutions, even if formally democratic, differ considerably in their ability to satisfy the grievances of groups within the society (Lijphart, 1977; Powell, 1982; Aksoy and Carter, 2011). As a consequence, groups who feel excluded from the political process and lack viable alternatives may ultimately resort to terrorist violence as a means to express their grievance and advance their political goals. Domestic versus Transnational Terrorism Terrorism represents a major threat to all states and human security, but not all countries and regions appear equally vulnerable, nor do they face the same type and level of risk. Although there already exists a significant body of research on terrorism, most existing systematic research on terrorism focuses exclusively on transnational terrorism, where either the perpetrators, venue, or victims or origins of attacks involve different countries, as opposed to domestic, home-grown, or home-directed terrorism within a single country (e.g. Enders and Sandler, 2008; Enders, Sandler and Gaibulloev, 2011; Savun and Phillips, Krieger and Meierrieks, 2010). Existing literature on 2

3 domestic terrorism is in fact primarily qualitative and often based on case-studies that only consider conflict cases and where the findings are difficult to generalise (Sanchez- Cuenca and de la Calle, 2009). The near exclusive transnational focus is unfortunate, as domestic terrorist attacks outnumber transnational attacks about four to one, and domestic attacks have generated a much larger number of deaths and victims (Enders and Sandler, 2006; La Free and Dugan, 2004; Savun and Phillips, 2009). Figure 1 and 2 present line plots of domestic and transnational terrorist attacks from 1970 to Clearly the number of domestic attacks is systematically larger than the number of transnational terrorist attacks, with peaks of 3000 and 800 attacks respectively. [Figure 1 about here] [Figure 2 about here] As a result of the focus on transnational terrorism we still know little about the causes of domestic terrorism, and how these may differ from transnational attacks. There are strong reasons to suspect that the explanatory frameworks for transnational terrorism are inadequate for understanding domestic terrorism. 3 If so, generalizing from research on transnational terrorism to domestic terrorism may yield misleading conclusions and flawed policy prescriptions for counterterrorism efforts. An example of such tendency to generalize from transnational to domestic terrorism is represented by the relationship between regime type and terrorism. It has been argued that in principle democracy may reduce the likelihood of terrorism because the presence of civil liberties, freedom of expression and multiple opportunities for political participation allow citizens to voice their grievances and lower the opportunity costs of achieving one s goals through legal means (e.g. Crenshaw, ). Most empirical studies, however, have supported the opposite relationship, namely, that democracy encourages terrorism. The argument is premised on the idea that democratic 3 To my knowledge, only five large-n studies have differentiated between domestic and transnational terrorism. These studies indicate that the determinants of these two forms of terrorism may indeed differ; see Piazza, 2011; Aksoy and Carter, 2011; Bloomberg and Hess, 2008; Sambanis, 2008; Savun and Phillips, Cfr. also Krieger and Meierrieks, 2010; Enders, Sandler and Gaibulloev, Although the author also acknowledges that democracies provide multiple opportunities for terrorist violence. 3

4 regimes constitute a permissive environment for terrorist activities since the necessity of adherence to civil liberties and other legal guarantees, such as freedom of association, make it easier for terrorist groups to organize and more difficult for the state to prosecute them (e.g. Eubank and Weinberg, 1994, 1998, 2001; Schmid, 1992). Furthermore the freedom of the press provides media coverage and publicity which help terrorist groups draw attention to their cause and attract more recruits. Finally, Li (2005) suggests that some features of democratic regimes, such as institutional constraints on governments, are likely to promote terrorism whereas other aspects, such as political participation, may actually reduce terrorist attacks. Yet, virtually all the studies which have investigated the relationship between regime type and terrorism have focused exclusively on transnational terrorism, which represents a rather small proportion of global terrorism. Moreover, these studies have generally adopted a monadic design which fails to account for the fact that multiple countries, with arguably different political characteristics, are affected by transnational terrorism either as target, origin or host and for the relationship between them. As a consequence, empirical results appear quite inconclusive, and in fact, as recent research has shown, the effect of democracy on transnational terrorism is not robust under dyadic specifications (Young and Findley, 2011). On the other hand, it remains unclear whether and how the characteristics of the political context matter for domestic terrorism. Young and Findley (2011) have demonstrated that when models of transnational terrorism are applied to domestic terrorism, results become inconsistent. Similarly, Piazza (2011) has introduced political participation and executive constraints in his economic model of domestic terrorism obtaining results which do not fully match those for transnational terrorism. Finally, Chenoweth (2010) finds that specific features of democratic regimes are not significant for explaining terrorism once we control for political competition. Given such puzzling results, extant research does not seem to provide a clear theoretical framework to understand the political determinants of domestic terrorism and how these may differ from transnational terrorism. 4

5 Determinants of Domestic Terrorism: A General Framework Unlike transnational terrorism, domestic terrorism is homegrown and home-directed: its origins and immediate effects are circumscribed to the host country (Rosendorff and Sandler, 2005:172). Therefore, to understand the country-level determinants of domestic terrorism it is important to look primarily within countries, and analyze how characteristics of the environment, that is, structural factors at the national and local level, affect the emergence of domestic terrorism. In fact, as it will be further argued, characteristics of the political context are not simply a background for group action but may affect groups motives and incentives to resort to terrorist violence. There are two fundamental ways in which structural factors may affect the likelihood of terrorism: they can influence terrorist groups motives or simply provide opportunities for terrorist actions. Accordingly, the country-level determinants of domestic terrorism can be divided into direct causes and permissive or enabling conditions, which will be referred to also as demand-side and supply-side factors. More specifically, direct causes or demand-side factors refer to grievances stemming from political, social and economic conditions which directly inspire and motivate terrorist activity (Crenshaw, 1981:381; Drakos and Gofas, 2006:75). Supply-side or permissive factors refer to conditions which provide opportunities for terrorism to occur (Crenshaw, 1981:381; Drakos and Gofas, 2006:75) in that they affect terrorists' strategic calculations by either maximizing the expected return or minimizing the costs of action. In this regard, existing quantitative research on the determinants of terrorism has not often distinguished between motives and opportunities: as previously mentioned, many studies have argued that a democratic regime type provides opportunities for transnational terrorism to occur; however Norway does not have the same likelihood of being targeted by transnational terrorism as the United States or the United Kingdom. The reason for this discrepancy may ultimately lie in the different motives for terrorist activity which those countries provide, for instance in the type of foreign policies they pursue (Savun and Phillips, 2009). 5

6 The Political Determinants of Domestic Terrorism This study intends to contribute to the research on determinants of terrorism by investigating whether and how the characteristics and quality of a country s political environment influence the likelihood of domestic terrorism. The focus will not be simply on regime type as an aggregate concept but on particular characteristics of regimes which affect their political openness the degree to which citizens are allowed to express their opinions and participate in the political process as well as how representative the political institutions are of different views and groups within the society. Recalling a famous definition by Bruce Hoffman, terrorism can be defined as the deliberate creation and exploitation of fear through violence or the threat of violence in the pursuit of political change (2006:40). In this regard, groups resort to terrorist violence in order to advance their political goals often in the absence of feasible alternatives (cfr. e.g. Crenshaw, 1981). In fact many terrorist groups have justified the use of terrorist violence by arguing that the political systems which they targeted was too restrictive for them to voice their grievances peacefully (Aksoy and Carter, 2011:2). This, in turn, suggests a relationship between a country s domestic political institutions and the outbreak of terrorist violence within that country. At the same time, however, political institutions, no matter how democratic, differ significantly in their ability to satisfy the grievances of societal groups (e.g. Powell, 1982). The concept of regime type as an aggregate of institutions and legal provisions appears ill-suited to capture this variation. Interestingly, the very few studies which have looked at institutional variation within formally similar regime types have lent support to the role of political grievances vs. opportunities in generating domestic terrorism (Aksoy et al ; Aksoy and Carter, 2011; Foster et al. 2011). In this study I argue that two crucial factors linking the political-institutional context with the emergence of domestic terrorist violence are the general openness of the political system and its degree of inclusiveness, that is, the extent to which different voices and groups within society are granted representation and access to power. In this regard it is important to notice that some institutional features of domestic politics affect 6

7 the population as a whole without significant communal differences whereas others predominantly affect specific sub-national groups. Political openness ensures citizens ability to participate freely in the civil and political life of a country, without discrimination or repression. In this respect, the level of civil liberties and political rights which is granted to individuals and groups is of particular importance. More democratic and politically open regimes allow citizens' access to non violent forms of political participation to pursue their interests and express their discontent and grievances (Savun and Phillips, 2009:881; Crenshaw, 1981; Ross, 1993). However, while fully democratic regimes may be less likely to experience domestic terrorism for the aforementioned reasons, highly authoritarian regimes may be so repressive as to deter or crack down any emerging dissent thus rendering violent action especially risky and difficult to organize. Therefore political openness may affect the likelihood of domestic terrorism in a non monotonic way: more specifically, countries characterized by intermediate levels of political rights and civil liberties as well as regimes undergoing a political transition may be more prone to domestic terrorism than other types of regime. H1: Countries with intermediate levels of civil liberties and political rights granted to their citizens are more likely to experience domestic terrorism. However, the degree of political openness per se does not tell us much about the representative inclusiveness of domestic institutions, that is, to what extent those institutions are capable of reflecting and even accommodating different political views, interests and even grievances of societal groups. In this regard, electoral systems represent a crucial determinant of the degree of inclusiveness of a political system (cfr. Lijphart, 1977; 1994). In fact, as Cohen (1997:612) argued, proportional electoral rules are more permissive for representation of different groups and interests within societies, enhance access to political institutions and decrease the stakes of competition. Additionally, more proportional systems provide institutional means for dissatisfied groups to voice their problems and even ethnic conflicts are less likely to escalate (Aksoy and Carter, 2011:6; Cohen, 1997). 7

8 It may be argued that electoral systems are not relevant outside democracies because, even though most countries in the world hold elections, in non democratic countries elections are not competitive, hence electoral systems do not matter. However, as recent research has shown (Hyde and Marinov, 2011) this is not necessarily the case. Competitive elections, namely, elections which can be lost by the party in office, where opposition is allowed, with more than one party and with a choice of candidates on the ballot, take place also in countries which are not fully democratic. In similar contexts electoral systems can play a significant role. As a consequence, and in contrast with existing approaches (e.g. Aksoy and Carter, 2011), it may be useful to discriminate cases in which electoral systems matter by looking at the competitiveness of the elections in which those electoral systems are applied rather than just focusing on democratic countries. This study departs from existing literature, based on categorical indicators of majoritarian, proportional and mixed electoral systems, and introduces a continuous measure of inclusiveness which combines information on the electoral formula and the district magnitude and places all existing electoral systems on a continuum from no inclusiveness (exclusive systems) to maximum inclusiveness (cfr. also Foster et al. 2011) H2: Countries with more inclusive electoral systems are less likely to experience domestic terrorism than countries with exclusive electoral systems. Looking at the electoral institutions to approximate the degree of inclusiveness of a political system may not be enough as this does not account for the existence of direct discriminatory policies whereby certain groups are systematically denied access to power. Accordingly, political discrimination of minorities and lack of access to power should also be considered. The latter, in fact, are indicative of segments of the population which do not enjoy the same rights and treatment as the majority of citizens, and do not have the same opportunities for political participation and representation. It also suggests that domestic political institutions, however democratic, may not be perceived as representative by part of the population and, especially within the context of a struggle for self-determination, they may even appear as symbols of an occupying 8

9 power (cfr. Pape, 2003, 2005). The quality of a country s political environment may not be homogenous in the presence of horizontal political inequalities. This, in turn, suggests the need to look not just at national-level characteristics but also at more localized conditions which may affect specific groups rather than the population as a whole. After all, as Piazza (2011:341) argued, terrorist groups, as small organized actors led by elites that draw recruits from aggrieved sub-national communities, are instruments of mobilization that allow group grievances to be channeled into violent activity. H3: Political discrimination of minority groups increases the likelihood of domestic terrorism. H4: Ethnic exclusion increases the likelihood of domestic terrorism. Research Design, Data, and Operationalisation (preliminary) The unit of analysis in the present study is country-year since the theory investigates how the characteristics of a country s political environment affect the likelihood of domestic terrorism. The dependent variable is a count of domestic terrorist attacks. Using the number of attacks as dependent variable is quite common in terrorism research as this reflects more accurately the magnitude of terrorism risk. Although previous research on terrorism has been hampered by a lack of data on domestic events, the new Global Terrorism Database (GTD) of the START programme at the University of Maryland provides systematic information on domestic and transnational terrorist events, including date, country and location of attacks, target types and nationality, group or individual responsible, for , and Enders, Sandler and Gaibulloev (2011) have devised procedures to separate domestic and transnational terrorist events. Accordingly GTD constitutes the main data source for secondary analysis. The estimation method is negative binomial regression due to overdispersion in the dependent variable. 9

10 Concerning the independent variables, political openness is operationalised using the Freedom House index of political rights and civil liberties. This variable ranges from 1 to 7, where 1 indicates maximum level (completely free regimes) and 7 the minimum (not free). Data on electoral systems is gathered from Golder s dataset on Electoral Systems Around the World (2004). I am currently in the process of coding the degree of inclusiveness of different electoral systems therefore this variable will not be included in the empirical analysis of the present draft. Information on political discrimination of minorities is obtained from the Minorities at Risk Project (University of Maryland). 5 This variable ranges from 0 (no discrimination) to 4 (Political exclusion and repressive policy). Moreover, to operationalise ethnic exclusion I create a variable which combines information on powerless and discriminated ethnic groups from the Ethnic Power Relations dataset (Cederman et al. 2009). The variable is dichotomous and takes the value of 1 for ethnic discrimination/exclusion and 0 otherwise. Preliminary control variables include GDP per capita, since richer countries are generally found to be more prone to terrorism (e.g. Bloomberg and Hess, 2008) and total population. Data for these variables are from Gleditsch (2002). I also control for whether a country experienced a civil war (Findley and Young, 2012). Additional control variables to be included in further specifications are political participation and competition (Li, 2005; Chenoweth, 2010), economic inequality (Piazza, 2011), and regional dummies. Following Carter and Signorino (2010), in all models I account for temporal dependence (2010) using a cubic polynomial in time. The temporal variable t measures the time since a country experienced domestic terrorist attacks. Including t, t 2, and t 3 in the models ensures that any temporal dependence in the number of terrorist attacks is explicitly controlled for. 5 Especially for ethno-nationalist/separatist terrorism country averages of political or economic inequalities may not be particularly significant (cfr. Piazza, 2011, and Derin-Güre, 2009). For a similar reasoning in the context of ethno-nationalist civil wars see Cederman, Weidmann and Gleditsch,

11 All models include robust standard errors clustered on the country level to account for the non independence of observations within countries. Empirical Results and Discussion (preliminary) Preliminary results of two negative binomial regressions on the number of terrorist attacks are presented in table 1. As previously mentioned, these models only test hypotheses one, three, and four since I do not currently have complete data on the inclusiveness of electoral systems. [Table 1 about here] The two models support the hypothesized curvilinear relationship between political rights and civil liberties on the one hand and the number of domestic terrorist attacks a country is likely to experience on the other. However, more substantive tests are needed to verify whether there is actually a turning point. Concerning the effect of political discrimination of minorities and ethnic exclusion, the hypothesized relationship is also borne out as their coefficients are positive and significant. Conclusion 11

12 0 Transnational Terrorist Attacks Domestic Terrorist Attacks Appendix Figure 1. Line plot of domestic terrorist attacks Domestic Terrorism Year Domestic Terrorist Attacks Figure 2. Line plot of transnational terrorist attacks Transnational Terrorism Year Transnational Terrorist Attacks 12

13 0 5,000 Number of Attacks 10,000 15,000 20,000 Figure 3. Distribution of domestic terrorist attacks by countries freedom status Terrorist Attacks by Country Status ( ) Free Not Free Partly Free 13

14 Table 1. Negative Binomial Regression Model 1 Model 2 Number of terrorist attacks Political rights t *** (0.0880) Political rights squared t *** (0.0109) Civil liberties t-1 Civil liberties squared t-1 Political discrimination of minorities Exclusion from power (ethnic groups only) *** *** (0.0887) *** (0.0108) *** (0.0227) (0.0229) *** *** (0.0724) (0.0721) GDP per capita ** ** ( ) ( ) Civil war *** *** (0.0811) (0.0832) Total population *** *** ( ) ( ) t *** *** (0.0365) (0.0365) t *** *** ( ) ( ) t *** *** ( ) ( ) Constant * (0.170) (0.182) lnalpha *** *** (0.0312) (0.0312) N * p < 0.05, ** p < 0.01, *** p <

15 Bibliography Abadie, A Poverty, Political Freedom, and the Roots of Terrorism. The American Economic Review, Vol. 96(2): Aksoy, D. and D. B. Carter, Electoral Institutions and the Emergence of Terrorist Groups. %20Carter.pdf Aksoy, D., D.B. Carter, and J. Wright Terrorism in Dictatorships, forthcoming in Journal of Politics. Beck, N., J. N. Katz, and R. Tucker Taking Time Seriously: Time-Series-Cross- Section Analysis with a Binary Dependent Variable. American Journal of Political Science Vol. 42: Bloomberg, S.B., and Hess, G.D From (No) Butter to Guns? Understanding the Economic Role in Transnational Terrorism, in Keefer, P., and Loayza, N. (eds.) Terrorism, Economic Development and Political Openness, New York: Cambridge University Press. Carter, D. B. and C. S. Signorino, Back to the Future: Modeling Time Dependence in Binary Data, Political Analysis, Vol. 18: Chenoweth, E Democratic Competition and Terrorist Activity, Journal of Politics, Vol.72(1): Cohen, Frank S Proportional Versus Majoritarian Ethnic Conflict Management in Democracies. Comparative Political Studies, Vol. 30: Crenshaw, M The Causes of Terrorism. Comparative Politics, Vol. 13(4): Cederman, L.E., N.B. Weidmann and K.S. Gleditsch Horizontal Inequalities and Ethno-nationalist Civil War: A Global Comparison, American Political Science Review, Vol. 105(3): Cederman, L.E., B. Min, A. Wimmer, 2009, "Ethnic Power Relations dataset", UNF:5:k4xxXC2ASI204QZ4jqvUrQ== V1 15

16 Derin-Güre, P Does Terrorism Have Economic Roots?, Unpublished Munuscript, Boston University. Drakos, K., and A. Gofas, In Search of the Average Transnational Terrorist Attack Venue. Defence and Peace Economics, Vol.17(2): Enders, W., and Sandler, T The Political Economy of Terrorism. Cambridge: Cambridge University Press. Enders, W., Sandler, T., and Gaibulloev, T Domestic versus Transnational Terrorism: Data, Decomposition and Dynamics, Journal of Peace Research, Vol. 48(3): Eubank, W., and L. Weinberg, Does Democracy Encourage Terrorism?. Terrorism and Political Violence, Vol. 6(4): Eubank, W. and L. Weinberg, Terrorism and Democracy: What Recent Events Disclose, Terrorism and Political Violence, Vol. 10(1): Findely, M. and J. Young Terrorism and Civil War: A Spatial and Temporal Approach to a Conceptual Problem, forthcoming in Perspectives on Politics. Foster, D., D. Sobek, and A. Braithwaite Tyranny of the Majority: Electoral Systems and Terrorism in Democracies. Working paper. Freedom House Reports, Freedom in the World, (accessed January 2012) Gleditsch, K. S Expanded Trade and GDP Data. Journal of Conflict Resolution 46: Version (accessed March 6, 2012). Golder, M Democratic Electoral Systems Around the World, Electoral Studies Vol. 24: GTD, Global Terrorism Database, < Hoffman, B Inside Terrorism, New York: Columbia University Press. Hyde, S. and N. Marinov, Which Elections Can be Lost? forthcoming in Political Analysis. Krieger, T., and Meierrieks, D What Causes Terrorism?. Public Choice. < 16

17 Krueger, A.B., and Laitin, D.D Kto Kogo?: A Cross-country Study of the Origins and Targets of Terrorism, in Keefer, P., and Loayza, N. (eds.) Terrorism, Economic Development and Political Openness, New York: Cambridge University Press. La Free, G., and Dugan, L How Does Studying Terrorism Compare to Studying Crime?, in Deflem, M., Terrorism and Counterterrorism: A Criminological Perspective, New York: Elsevier. Li, Q Does Democracy Promote or Reduce Transnational Terrorist Incidents?. Journal of Conflict Resolution, Vol.49(2): 1-20 Lijphart, A Democracy in Plural Societies: A Comparative Exploration. New Haven: Yale University Press. Lijphart, A Electoral Systems and Party Systems, New York: Oxford University Press. Minorities at Risk Project, "Minorities at Risk Dataset." College Park, MD: Center for International Development and Conflict Management. Pape, R 'The Strategic Logic of Suicide Terrorism' American Political Science Review, 97 (3): Pape, R Dying to Win. The Strategic Logic of Suicide Terrorism. New York: Random House. Piazza, J Poverty, Minority Economic Discrimination and Domestic Terrorism, Journal of Peace Research, Vol. 48(3): Powell, G. Bingham Contemporary Democracies, Participation, Stability, and Violence, Cambridge, Mass: Harvard University Press. Rosendorff, P.B., and Sandler, T The Political Economy of Transnational Terrorism. Journal of Conflict Resolution, Vol.49: Ross, J.I Structural Causes of Oppositional Political Terrorism: Towards a Causal Model. Journal of Peace Research, Vol.30(3): Sambanis, N Terrorism and Civil War, in Keefer, P., and Loayza, N. (eds.) Terrorism, Economic Development and Political Openness, New York: Cambridge University Press. 17

18 Sandler, T., and Enders, W Economic Consequences of Terrorism in Developed and Developing Countries, in Keefer, P., and Loayza, N. (eds.) Terrorism, Economic Development and Political Openness, New York: Cambridge University Press. Savun, B., and Phillips, B Democracy, Foreign Policy, and Terrorism. Journal of Conflict Resolution, Vol.53(6): Sobek, D., and Braithwaite, A Victim of Success: American Dominance and Terrorism. Conflict Management and Peace Science, Vol START, National Consortium for the Study of Terrorism and Responses to Terrorism < Tilly, C The Politics of Collective Violence. Cambridge: Cambridge University Press. Young, J. and M. Findley, Promise and Pitfalls of Terrorism Research, International Studies Review, Vol. 13: